Low-Energy Electron-Stimulated Luminescence of Thin H2O and D2O

Fundamental Science Directorate, Pacific Northwest National Laboratory, Mail Stop K8-88, P.O. Box 999, Richland, Washington 99352. J. Phys. Chem. B , ...
0 downloads 0 Views 109KB Size
Download PDF
J. Phys. Chem. B 2005, 109, 15835-15841

15835

Low-Energy Electron-Stimulated Luminescence of Thin H2O and D2O Layers on Pt(111) Nikolay G. Petrik and Greg A. Kimmel* Fundamental Science Directorate, Pacific Northwest National Laboratory, Mail Stop K8-88, P.O. Box 999, Richland, Washington 99352 ReceiVed: December 29, 2004; In Final Form: April 11, 2005

The electron-stimulated luminescence (ESL) from amorphous solid water and crystalline ice films deposited on Pt(111) at 100 K is investigated as a function of the film thickness, incident electron energy (5-1000 eV), isotopic composition, and film structure. The ESL emission spectrum has a characteristic double-peaked shape that has been attributed to a transition between a superexcited state (C ˜ ) and the dissociative, first excited state (A ˜ ) in water: C ˜ fA ˜ . Comparing the electron-stimulated luminescence and O2 electron-stimulated desorption (ESD) yields versus incident electron energy, we find the ESL threshold is ∼3 eV higher than the O2 ESD threshold, which is close to the center of the emission spectrum near 400 nm and supports the C ˜ fA ˜ assignment for the ESL. For thin films, radiative and nonradiative interactions with the substrate tend to quench the luminescence. The luminescence yield increases with coverage since the interactions with the substrate become less important. The ESL yield from D2O is ∼4-5 times higher than that from H2O. With use of layered films of H2O and D2O, this sizable isotopic effect on the ESL is exploited to spatially profile the luminescence emission within the ASW films. These experiments show that most of the luminescence is emitted from within the penetration depth of the incident electron. However, the results depend on the order of the isotopes in the film and can be modeled by assuming some migration of the electronically excited states within the film. The ESL is very sensitive to defects and structural changes in solid water, and the emission yield is significantly higher from amorphous films than from crystalline ice.

I. Introduction High-energy radiation (γ-rays, R particles, X-rays, etc.) produces cascades of low-energy secondary electrons in condensed matter, and the subsequent reactions of these low-energy electrons cause most of the chemical and physical changes in aqueous systems. The mechanisms of these electron-driven processes in water have been characterized in research over the last several decades.1 More recently, the interactions of condensed water with low-energy electrons2-8 have been studied and a variety of nonthermal reaction mechanisms have been identified. There are three major primary processes that can cause chemical changes in condensed water under electron irradiation: ionization, direct excitation, and dissociative electron attachment.3,5,9-11 The primary processes and the final “bondbreaking” events can be well separated in space. Recent experiments6,8 have shown that electron-stimulated reactions in thin ASW films occur preferentially at or near the ASW/vacuum and Pt/ASW interfaces. However, the energy driving these reactions can migrate through the films over long distances, e.g. up to ∼150 nm corresponding to 400 monolayers (ML). The mechanisms for the energy migration are not well understood. Similar energy transfer was observed in alkali halides and rare gas solids, where the location of an initial electron excitation and the subsequent reactions can also be spatially separated.12-14 In those cases, excitons created in the bulk can diffuse to the vacuum interfaces where they subsequently lead to desorption, and a similar mechanism might occur in water. Investigating the nonthermal luminescence can also provide insight into the mechanisms of energy transfer and relaxation in aqueous systems. For example, the radioluminescence of * Address correspondence to this author. E-mail: [email protected].

crystalline water ice has been studied with 50 keV X-rays,15-19 high-energy (>0.5 MeV) electrons,20-25 and low-energy ( 10 eV).30 Potential energy surface calculations suggested that the main part of the emission spectra is due to the C ˜ fA ˜ electron transition from a highly excited state to the dissociative, first excited state of water:30

H2O(X ˜ ) + pω f H2O(C ˜ ) f H2O(A ˜ ) + pω f H + OH(X) The complicated shape of the gas-phase emission spectrum is the result of several vibrational states of the OH, which effect the partial emission cross-sections.30 The emission spectra for amorphous solid water and crystalline ice are blue shifted by

Electron-Stimulated Luminescence of Water Films

Figure 2. Integrated ESL emission yield (open circles) and O2 ESD yield (filled circles) versus energy of incident electrons for 150 ML D2O films deposited at 100 K. The data show onsets at 11.5 ( 0.5 eV for ESL and at 8.5 ( 0.5 eV for O2 ESD.

∼50 nm compared to the gas-phase emission spectrum probably due to matrix effects on the C ˜ fA ˜ transition.23 The H2O and D2O emission spectra look similar within our sensitivity and resolution limits (Figure 1), but the ESL yield from H2O is ∼3 times lower than that from D2O. (As discussed below, the initial ESL yield from H2O is ∼4-5 times smaller. However, the H2O ESL decays less with electron dose, and the spectra in Figure 1 are obtained after the ESL has decayed to its saturation value.) The lower yield for H2O agrees with earlier observations for crystalline ices, and likely relates to more efficient quenching of excited states in H2O via vibrational relaxation.19,27 If the ESL is primarily associated with the C ˜ fA ˜ transition in water, then the ESL threshold should occur at higher incident electron energy than the threshold for processes related to excitation of the lowest excited state. Figure 2 shows total ESL emission yield (i.e., measured without the monochromator) versus the incident electron energy. For these data, 150 ML D2O films were deposited and irradiated at 100 K. The ESL yield has an apparent threshold at 11.5 ( 0.5 eV, which is close to the 11-12 eV thresholds reported earlier,26,27 and similar to 11.4 eV estimated for X ˜ fC ˜ transition in ice.23 We have also measured the O2 ESD yield versus incident electron energy for 150 ML D2O films. In this case, the apparent threshold is at 8.5 ( 0.5 eV. It is believed that the onset for the O2 ESD corresponds to excitation of the lowest excited state in solid water,35,36 which was measured in the ASW at 8.5 eV by EELS.37 The ∼3 eV shift between the ESL and ESD thresholds is very close to the photon energy at the maximum of the luminescence spectra (∼400 nm), which supports the emitter assignment as the superexcited C ˜ state. As discussed below, it is difficult to measure the ESL threshold versus electron energy from very thin ASW layers since the luminescence yield decreases dramatically. For thicker films (e.g. 150 ML), the measured threshold value may experience some systematic shift because of electron-induced charging of the film. However, decreasing the incident electron beam current and decreasing the film thickness to 100 ML did not change the apparent threshold within the uncertainty of the measurement. We believe the total charging is small primarily due to the fact that the ESD yields of D2O and D2 are constant versus time for thick ASW films (data not shown). Above the threshold, the luminescence yield monotonically increases with increasing electron energy up to the highest energies investigated (Ei ) 80 eV, data not shown). The increase of the ESL yield with electron energy indicates that in addition to direct excitation, ionization followed by electron-ion recombination also leads to formation of the excited states responsible for the luminescence. Since the luminescence comes from a superexcited state, the electron-ion recombination must

J. Phys. Chem. B, Vol. 109, No. 33, 2005 15837

Figure 3. Integrated ESL yield versus ASW coverage for both D2O (filled circles) and H2O (open squares). The ESL yield was integrated during the first 10 s of irradiation. The luminescence yields are low for coverages ∼22 ML. Conversely, for D2O deposited on top (open squares), the ESL signal increases with increasing D2O coverage and saturates for θD2O > ∼35 ML. These results show that the ESL is localized in the outer 2030 ML of the ASW films, depends on the isotope present in that region of the film, and is independent of the isotopic composition deeper within the film. In Figure 4b, we show the ESL yield from 30 ML of D2O deposited on top of various amounts of H2O versus the total coverage of the water film.

15838 J. Phys. Chem. B, Vol. 109, No. 33, 2005

Petrik and Kimmel

Figure 5. Integrated ESL yield from the isotopically layered films Pt/H2O/D2O versus the D2O cap layer coverage, θD2O, for 87 eV incident electrons (filled circles) and for the 1000 eV incident electrons (open squares) where the total water coverage, θT, is constant: θT ) θD2O + θH2O ) 155 and 250 ML, respectively. The results indicate that most of the ESL is emitted from the outer layer, close to the penetration depth of the incident electrons: ∼10 and ∼50 nm for 87 and 1000 eV electrons, respectively (taking into account the incident angle of the electrons).

Figure 4. (A) Integrated ESL yield from the isotopically layered films Pt/H2O/D2O (open squares) and Pt/D2O/H2O (filled circles) versus the coverage of the D2O or H2O cap layer where the total water coverage, θT, is constant, θT ) θD2O + θH2O ) 155 ML (i.e., the fraction of each isotopic layer is changed from 0 to 1). The lines are the results of a model discussed in the text (see also Figure 6). (B) ESL yield from a 30 ML of D2O layer on top of an H2O spacer versus the total water coverage (open squares). For reference, the ESL yields from pure D2O films are also shown (filled circles). Both ESL yields versus coverage are practically identical.

Also shown for comparison is the ESL from pure D2O versus D2O coverage (from Figure 3). The ESL signal from 30 ML of D2O on an H2O spacer layer (squares) is essentially identical with the ESL signal from pure D2O (circles) indicating that, for all coverages, the ESL emission is localized in the top ∼20-30 ML of the ASW films. The results in Figure 4 indicate that for 87 eV electrons most of the electron-stimulated luminescence originates from the outer 20-30 ML of the ASW films. In crystalline ice at 100 K, the distance between successive planes along the c-axis is 0.366 nm.39 By using this distance for the ASW films to convert from coverage to thickness and taking into account the 45° angle of incident for the electrons, 20-30 ML corresponds to an electron penetration depth of ∼10-15 nm. This distance is close to the penetration depth of ∼5-10 nm calculated for 100 eV electrons in condensed water.40 In this case, the “penetration depth” refers to the region where most of the electronic excitations are produced within the film. Since the films do not charge appreciably (vide supra), subexcitation electrons and/or mobile holes must migrate through the film.41 Increasing the electron energy increases the penetration depth and increases the thickness of the layer from which the ESL is emitted: The integrated ESL yield from isotopically layered 250 ML ASW films irradiated with 1000 eV electrons increases versus the D2O cap layer coverage (squares) until θD2O ∼ 100 ML, and is constant at higher coverages (Figure 5). These data imply a penetration depth of ∼50 nm for the incident electrons, which is close to the calculated value for 1000 eV electrons in condensed water.40 In Figure 5, the results for 87 eV electrons irradiating a D2O cap layer (i.e., Figure 4a) are also shown for comparison (circles).

Figure 6. Integrated ESL yield from the isotopically layered films Pt/H2O/D2O (open squares) and Pt/D2O/H2O (filled circles) versus the coverage of the D2O or H2O cap layer where the total amount of ASW is constant (see Figure 4a). The lines show the results of modeling the ESL with energy transfer (solid lines) and without energy transfer (dashed lines) from the D2O to the H2O layers (see text for details).

Since the ESL from 87 eV electrons occurs only in the outer ∼20-30 ML of a water film (Figure 4), we need to consider why the ESL yield increases with increasing coverage up to ∼240 ML (Figure 3). It is likely that the rest of the film underneath the emitting layer serves as a dielectric spacer separating the emitting dipoles from the metal substrate and affecting the luminescence yield. In fact, similar distancedependent fluorescence yields and lifetimes have been observed in many other metal-dielectric systems.42,43 There are two main processes affecting the luminescence yield near metal surfaces: 42,43 First when the distance between an excited atom or molecule and the surface is considerably less than the wavelength of the emitted radiation, then nonradiative transfer of energy from the excited molecule to the metal tends to quench the luminescence. The nonradiative transfer is similar to dipoledipole (Fo¨rster) energy transfer.43 Second, destructive interference effects with the light reflected from the substrate reduce the luminescence for thicknesses comparable to the wavelength of the light. The experiments with layered films of H2O and D2O show that the excited states responsible for the ESL do not migrate over distances that are large compared to the penetration depth of the incident electrons (Figure 4). However, these same experiments show an asymmetry depending on which isotope is deposited on top that is probably due to energy transfer or migration on a shorter length scale (Figure 6). Three features of the data highlight the asymmetry: First, the magnitude of

Electron-Stimulated Luminescence of Water Films the initial slope of the luminescence signal versus cap layer coverage is different for the two cases. In particular, when H2O is on top, the luminescence decreases approximately twice as quickly compared to the rate of increase when D2O is on top. Second, the luminescence yield at the point where the data curves cross in Figure 6 is less than the average of the luminescence yields for pure D2O and H2O, i.e., the curves cross at ∼0.43 instead 0.6. Third, for H2O on top the luminescence saturates θH2O > ∼22 ML versus saturating at θD2O > ∼35 ML for D2O on top. If the isotopes were equal except for the luminescence yield (i.e., symmetric), the magnitude of the initial slopes would be the same, the luminescence curves would cross at the average value of yield for the pure films, and the luminescence would saturate at the same coverages (dashed lines, Figure 6). The spatial distribution of electronic excitations is probably independent of the isotope.44,45 Therefore, the data suggest that electronic excitations created in the D2O layer can migrate to the H2O layer and subsequently decay nonradiatively. To model the data, we can use a simple one-dimensional Monte Carlo simulation that takes into account the initial spatial distribution of the excitations, their subsequent diffusion, and luminescence or quenching. The energy deposition per unit length of primary electrons (i.e., electronic excitations and ionizations) in the film can be approximated by a Gaussian I(x) ) dE/dx ) exp[-(x x0)2/2σ2], where σ is the standard deviation and I(x) is maximized at x0.40 To run a simulation, a cap layer thickness and isotope ordering (i.e., H2O or D2O on top) are specified. An initial location for an excitation is randomly chosen from the Gaussian distribution. For each Monte Carlo iteration, the excitation either takes a random walk step (i.e., diffuses), luminesces, or quenchs. So Pdiff + PL + PQ ) 1, where Pdiff, PL, and PQ are the probabilities for diffusion, luminescence, and quenching, respectively. Pdiff, PL, and PQ depend on the isotope at the current position of the excitation. Each excitation “trajectory” is followed until it luminesces or quenchs, and excitation “trajectories” are run until sufficiently good statistics are obtained for the luminescence yield for this configuration. The procedure is then repeated for each cap layer thickness and isotope ordering, to model the data in Figure 6. The model has several parameters that need to be chosen. First, we use calculations of dE/dx for 100 eV electrons in water,40 to determine σ and x0: σ ) 6 layers (∼2.2 nm) and x0 ) 0. Note that almost any distribution with a similar length scale for the penetration depth of the beam will give qualitatively similar results. Second, the experimental ESL yield from pure D2O is approximately 5 times greater than the ESL yield from pure H2O (see Figures 3 and 4 for large cap layer coverages). This determines the ratio, RLQ, of the luminescence to quenching probabilities for each isotope: RLQ(D2O) ) 5:1 and RLQ(H2O) ) 1:5. Therefore, since the probabilities must sum to one, there is only one independent probability for each isotope, and the model has two adjustable parameters. We adjust the diffusion probabilities, Pdiff(H2O) and Pdiff(D2O), and express them in terms of the average number of random walk steps, Nsteps(H2O) and Nsteps(D2O), an excitation in a pure film would make prior to either luminescing or quenching: Nsteps ) (1 - Pdiff)-1 - 1. The solid lines in Figure 6 show the results of a simulation for Nsteps(H2O) ) 100 and Nsteps(D2O) ) 300, which qualitatively reproduces the data including the initial slopes, crossing point, and saturation values discussed above. The simulation suggests that for the pure D2O (H2O) films the typical distance (normal to the surface) between where an excitation is created and where it either emits or quenches is ∼6 nm (∼3.7 nm) corresponding

J. Phys. Chem. B, Vol. 109, No. 33, 2005 15839 to ∼17 ML (∼10 ML). For comparison, the dashed lines show the simulation results for the “symmetric” case where Nsteps(H2O) ) Nsteps(D2O) ) 100. As noted above, the σ used in the simulations is based on published reports. However, comparable fits to the data can be obtained for simulations with a larger penetration depth for the incident electrons (i.e., larger σ) and less diffusion of the electronic excitation. Therefore, the data and simulation do not allow us to unambiguously determine these two parameters. Other experiments measuring various electron-stimulated reactions versus coverage6,8,46 suggest a somewhat larger penetration depth corresponding to σ ∼ 8-10 layers. For simulations with σ in this range, the typical diffusion distances are somewhat smaller (∼8-15 layers, depending on the isotope). Energy transfer from D2O to H2O is consistent with X-ray radioluminescence experiments on homogeneous mixtures of H2O and D2O.19 In those experiments, the luminescence yield was not proportional to the mole fraction-weighted yield of each isotope. Instead, the luminescence yield of isotopically mixed ice films was similar to the yield from pure H2O films even for low concentrations of H2O. Furthermore, the radioluminescence yield (at ∼400 nm) of crystalline ice is significantly enhanced by the addition of some impurities in low concentrations (e.g., alkali halide salts at ∼10-3 mol/L).15,19 The impurities may produce structural defects in the crystalline ice that promote luminescence. The observation that several different impurities at low concentrations affect the radioluminescence yields in similar ways is consistent with the hypothesis that the C ˜ -excitons are mobile. The luminescence is primarily due to the emission from an electronically excited state of water (the C ˜ state) that is formed either by direct excitation from the ground state or by electron-ion recombination. If the C ˜ state is formed by electron-ion recombination, the recombination must happen on a time scale that is short compared to the relaxation time for an energetic electron in the excited-state manifold (e.g., the conduction band in crystalline ice), otherwise the resulting excited state would be lower in energy. The short time available would seem to preclude significant hole hopping. Therefore for both direct excitation and electron-ion recombination to form the C ˜ state, it is probably the C ˜ state that is mobile and is responsible for the energy transfer from the D2O to the H2O films. The experiments with layered films of H2O and D2O are insensitive to diffusion parallel to the surface. Therefore, a typical diffusion distance of 10 ML in one dimension should correspond to several hundred random hops for the exciton in three dimensions. (In the amorphous films there is no reason to expect a preferred direction for the excitons to diffuse.) The relatively short-range migration of the C ˜ -excitons responsible for the ESL does not correspond to the much longer distances for the energy migration observed in the ESD experiments where energy transfer to the buried Pt/water interface can be observed over several hundred monolayers.8 Therefore we do not believe that these highly excited states are involved in such long-range energy transfer. Nonetheless, the mobility of these high-energy excitons is noteworthy. Excitons in solids diffuse via correlated hopping of their electrons and holes. The exciton’s range depends on the lifetime of the exciton and on its mobility. We expect the rate-limiting step to be the hole hopping step since the electronic wave functions associated with the hole are lower in energy and therefore more localized. The lowest energy exciton (corresponding to the lowest energy excited state in the gas phase) should have a longer lifetime

15840 J. Phys. Chem. B, Vol. 109, No. 33, 2005

Figure 7. Integrated ESL yield versus coverage for amorphous (open circles) and crystalline (filled squares) films of H2O on Pt(111). The amorphous (crystalline) films were grown at 100 K (140 K). The ESL yield from the crystalline films is ∼6-7 times lower than the ESL yield from the amorphous samples.

than the C ˜ -excitons. Therefore, since they share the same hole state,47 it is reasonable to expect that the lower energy exciton can migrate further. Figure 7 shows the ESL integrated yields versus coverage for amorphous (open circles) and crystalline ice (filled squares) films of H2O on Pt(111). The ESL yield from the crystalline ice films is ∼6-7 times lower than the ESL yield from the amorphous samples. Similar results are obtained for amorphous and crystalline D2O films (data not shown). There are several possible explanations for the large differences in the ESL between crystalline ice and ASW. One possibility is that the excited electrons scatter less in the crystalline films thus traveling further from the hole and suppressing exciton formation via electron-ion recombination. The electron-stimulated reaction yields in water films, which involve electron-ion recombination, also depend on the phase (crystalline or amorphous) although the effect is not as strong. However, we believe that the ESD of neutral products such as atomic and molecular hydrogen and oxygen primarily involves lower energy excitons. In that case, the electron-ion recombination step might not be as sensitive to the phase. Another possibility is that the luminescence happens in the vicinity of some structural defects in the solid water network that has a higher concentration in the ASW films. The radioluminescence yields in crystalline ice increase versus radiation fluence for 50 keV X-ray irradiation15,18,19 and 0.5 MeV electron irradiation.20 In these cases, the higher energy radiation may create defects in the crystalline ice that enhance the luminescence. In our experiments on ASW, the ESL yield from ASW decreases with increasing irradiation fluence. Figure 8 shows the ESL yield from 150 ML of amorphous D2O and H2O versus irradiation time. The ESL yield increases promptly when the electron beam is turned on, reaches a maximum after ∼0.5 s, and then slowly decays approaching a steady state value on a scale of minutes (for the given beam current). After the e-beam is turned off, the ESL signal decays promptly to the base level.48 If the irradiation is resumed, the ESL signal quickly returns to the level achieved at the end of the previous irradiation cycle. The normalized emission dose-dependence does not change with the ASW film thickness, but the ESL declines more slowly with increasing electron dose for H2O (Figure 8, inset). The last two results suggest that charging of the ASW films during irradiation is not responsible for the ESL dosedependence. Since the geometrical and electronic structures of the sites within the films where luminescence occurs are not known, a detailed explanation of the results in Figures 7 and 8 is not possible. One possibility is that, unlike high-energy irradiation of crystalline ice, low-energy electron irradiation of

Petrik and Kimmel

Figure 8. ESL yield versus irradiation time for 150 ML of amorphous D2O and H2O. The normalized emission dose-dependence is shown in the inset.

ASW “anneals” the film, removing defects that may be associated with the luminescence. In support of this, temperature programmed desorption spectra show that irradiated ASW films crystallize earlier in the temperature ramp than unirradiated films (data not shown). However, the ionizations and electronic excitations produced by the energetic electrons also create various reaction products (i.e., defects) in the ASW films. The spatial distribution of these radiolysis products within the film and their relationship to the ESL is unclear. For example, recent experiments suggest that low-energy, neutral excited states may dissociate only at the interfaces of the ASW films.6,8 Since the ESL originates from within the penetration depth of the energetic electrons, reaction products accumulating at the interfaces may have only a small influence on the ESL. However, higher energy excitations may lead to dissociation within the films and the creation of defects. More research is necessary to identify the sites associated with the luminescence and the effects of various types of radiation on these sites. IV. Conclusions In summary, we have studied the low-energy electronstimulated luminescence of thin films of D2O and H2O deposited on Pt(111). We find that the emission spectrum is centered at ∼400 nm and the luminescence has a threshold at ∼11.5 eV incident electron energy. The results are consistent with the assignment of the luminescence to the electronic transition in water molecule from the ground state to the superexcited state X ˜ fC ˜ , followed by a radiative transition to the lowest excited state C ˜ fA ˜ . The C ˜ state is formed by both direct excitation and electron-ion recombination. The luminescence yield is suppressed by the metal substrate and increases with distance from the metal in the coverage range 0-200 ML. The electronstimulated luminescence is emitted from the thin layer of the film at the vacuum side limited by the penetration of the incident electrons, corresponding to ∼10 and ∼50 nm for 87 and 1000 eV electrons, respectively. The electronic excitations responsible for the observed luminescence can migrate over a modest distance (∼10 ML in H2O) as shown by isotopic layering experiments. The emission yields from the amorphous solid water films are significantly higher than those from the crystalline water films, which may indicate involvement of structural defects in the emission process. Acknowledgment. This work was supported by LDRD investments from PNNL. Pacific Northwest National Laboratory

Electron-Stimulated Luminescence of Water Films is operated for the U.S. Department of Energy by Battelle Memorial Institute under contract DE-AC06-76RLO 1830. References and Notes (1) Farhataziz; Rodgers, M. A. J. Radiation chemistry: principles and applications; VCH Publishers: New York, 1987. (2) Noell, J. O.; Melius, C. F.; Stulen, R. H. Surf. Sci. 1985, 157, 119. (3) Rowntree, P.; Parenteau, L.; Sanche, L. J. Chem. Phys. 1991, 94, 8570. (4) Kimmel, G. A.; Orlando, T. M. Phys. ReV. Lett. 1995, 75, 2606. (5) Kimmel, G. A.; Orlando, T. M. Phys. ReV. Lett. 1996, 77, 3983. (6) Petrik, N. G.; Kimmel, G. A. Phys. ReView Lett. 2003, 90, 166102. (7) Petrik, N. G.; Kimmel, G. A. J. Chem. Phys. 2004, 121, 3727. (8) Petrik, N. G.; Kimmel, G. A. J. Chem. Phys. 2004, 121, 3736. (9) Kimmel, G. A.; Orlando, T. M.; Vezina, C.; Sanche, L. J. Chem. Phys. 1994, 101, 3282. (10) Kimmel, G. A.; Tonkyn, R. G.; Orlando, T. M. Nucl. Instrum. Methods Phys. Res., Sect. B 1995, 101, 179. (11) Orlando, T. M.; Kimmel, G. A.; Simpson, W. C. Nucl. Instrum. Methods Phys. Res., Sect. B 1999, 157, 183. (12) Song, A. K. S. Self-trapped Excitons; Springer-Verlag: Berlin, Germany; Heidelberg GmbH & Co. KG: Berlin, Germany, 1996. (13) Reimann, C. T.; Johnson, R. E.; Brown, W. L. Phys. ReV. Lett. 1984, 53, 600. (14) Shi, H.; Cloutier, P.; Gamache, J.; Sanche, L. Phys. ReV. B: Condens. Matter 1996, 53, 13830. (15) Grossweiner, L. I.; Matheson, M. S. J. Chem. Phys. 1954, 22, 1514. (16) Grossweiner, L. I.; Matheson, M. S. J. Chem. Phys. 1952, 20, 1654. (17) Ghormley, J. A. J. Chem. Phys. 1956, 24, 1111. (18) Steen, H. B. Chem. Phys. Lett. 1975, 35, 508. (19) Steen, H. B.; Holteng, J. A. J. Chem. Phys. 1975, 63, 2690. (20) Quickenden, T. I.; Trotman, S. M.; Sangster, D. F. J. Chem. Phys. 1982, 77, 3790. (21) Quickenden, T. I.; Matich, A. J.; Bakker, M. G.; Freeman, C. G.; Sangster, D. F. J. Chem. Phys. 1991, 95, 8843. (22) Freeman, C. G.; Quickenden, T. I.; Litjens, R. A. J.; Sangster, D. F. J. Chem. Phys. 1984, 81, 5252. (23) Vernon, C. F.; Matich, A. J.; Quickenden, T. I.; Sangster, D. F. J. Phys. Chem. 1991, 95, 7313. (24) Trotman, S. M.; Quickenden, T. I.; Sangster, D. F. J. Chem. Phys. 1986, 85, 2555.

J. Phys. Chem. B, Vol. 109, No. 33, 2005 15841 (25) Miyazaki, T.; Nagasaka, S.; Kamiya, Y.; Tanimura, K. J. Phys. Chem. 1993, 97, 10715. (26) Merkel, P. B.; Hamill, W. H. J. Chem. Phys. 1971, 54, 1695. (27) Prince, R. H.; Sears, G. N.; Morgan, F. J. J. Chem. Phys. 1976, 64, 3978. (28) Bernas, A.; Truong, T. B. Chem. Phys. Lett. 1974, 29, 585. (29) Yada, T.; Norizawa, K.; Hirai, M.; Yamanaka, C.; Ikeya, M. Jpn. J. Appl. Phys., Part 1 2002, 41, 5874. (30) Engel, V.; Mejer, G.; Bath, A.; Anderson, P.; Schinke, R. J. Chem. Phys. 1987, 87, 4310. (31) Stevenson, K. P.; Kimmel, G. A.; Dohnalek, Z.; Smith, R. S.; Kay, B. D. Science 1999, 283, 1505. (32) Kimmel, G. A.; Stevenson, K. P.; Dohnalek, Z.; Smith, R. S.; Kay, B. D. J. Chem. Phys. 2001, 114, 5284. (33) Dohnalek, Z.; Kimmel, G. A.; Ayotte, P.; Smith, R. S.; Kay, B. D. J. Chem. Phys. 2003, 118, 364. (34) Jones, W. M. J. Chem. Phys. 1952, 20, 1974. (35) Orlando, T. M.; Sieger, M. T. Surf. Sci. 2003, 528, 1. (36) Sieger, M. T.; Simpson, W. C.; Orlando, T. M. Nature 1998, 394, 554. (37) Michaud, M.; Cloutier, P.; Sanche, L. Phys. ReV. A 1991, 44, 5624. (38) Reimann, C. T.; Brown, W. L.; Johnson, R. E. Phys. ReV. B 1988, 37, 1455. (39) Petrenko, V. E.; Whitworth, R. W. Physics of Ice; Oxford University Press Inc.: New York, 1999. (40) LaVerne, J. A.; Pimblott, S. M. J. Phys. Chem. A 1997, 101, 4504. (41) Prince, R. H. Phys. Status Solidi B 1976, 78, 271. (42) Chance, R. R.; Prock, A.; Sibley, R. AdV. Chem. Phys. 1978, 37, 1. (43) Waldeck, D. H.; Alivisatos, A. P.; Harris, C. B. Surf. Sci. 1985, 158, 103. (44) Jonah, C. D.; Chernovitz, A. C. Can. J. Phys. 1990, 68, 935. (45) Experiments indicate that the distance subexcitation energy electronss those with energies below the threshold for electronic excitationsstravel does depend on the isotope (ref 44). (46) Petrik, N. G.; Kimmel, G. A. J. Chem. Phys. 2004, submitted for publication. (47) Tsurubuchi, S. Chem. Phys. 1975, 10, 335. (48) A significant part of the signal rise/fall time in these experiments is associated with the electron-gun’s current settling time (∼0.2 s).